By Harahsheh, Y., Duff, O.C., Ho, K.M. “Thromboelastography predicts thromboembolism in critically ill coagulopathic patients.” CCM. 2019;47(6): 826-832.
Summary: Thromboembolism is the main cause of preventable hospital-associated death. Coagulopathy is common in the ICU, often resulting in cessation of venous thromboembolism (VTE) prophylaxis, but some of these patients with abnormal coagulation may remain at high risk of VTE. This prospective cohort aimed to use thromboelastography (TEG) to risk-stratify risk of VTE in coagulopathic patients admitted to all ICUs at a tertiary referral center, excluding patients in the burn ICU, CTICU and transplant ICUs. Included patients had at least one abnormal coagulation test, defined as INR>1.5, aPTT>40s, or platelets<150x109/L within 48 hours of admission. TEG was measured using a single blood sample taken within 48 hours of admission. The primary outcome was any symptomatic thromboembolism and secondary outcome was transfusion requirements within 24 hours of admission. Other possible thrombotic markers also were assessed including CAT for global thrombin generation assessment, prothrombin fragment quantification, thrombin-antithrombin complex quantification, procoagulant microparticles and plasma P-selectin concentrations. Power analysis was done and demonstrated a need for a sample size of 200 to achieve 80 percent power to identify a AUROC greater than 0.70. Sensitivity analyses were performed to determine if maximum amplitude (MA) predicted thromboembolism differently between three groups: (1) arterial vs. venous events, (2) trauma vs. non-trauma and (3) after propensity score adjustment.
Results: Of 215 patients included in the final analysis, 34 (15.8 percent) developed symptomatic thromboembolism: 23 were venous (PE or DVT) and 11 were arterial. The median age did not differ significantly between the two groups (52 vs. 56). Lab measures of coagulation status were not significantly different between the two groups, although the platelet group was slightly lower in the control group (94K, IQR 67-128) compared to the case group (117K, IQR 80-145) with a p-value of 0.056. Median INR was 1.4 (IQR 1.2-1.7) in both groups and aPTT was 36.8s (32.1-46.2) in the case group compared to 40.6s (34.6-46.4) in the control. Fibrinogen was 4.1g/L (3.1-5.7) in cases compared to 3.7g/L (2.6-5.2). In the case group, 76.5 percent of patients were on pneumatic calf pumps compared to 82.3 percent in the non-thromboembolic events group; 44.1 percent were on unfractionated heparin compared to 44.2 percent in the non-thromboembolic group. Incidence of thromboembolism was significantly different based on reduced, normal or increased MA (P<0.001). Median MA for the thrombotic group was 71.7 (65.1-74.2) vs. 62.2 (53.8-68.7) in the control group (p<0.001). Other TEG parameters that also were significantly different between the two groups included clot kinetics (p=0.034) and alpha angle (p=0.010). The MA discriminated between patients with and without thromboembolism with an AUROC of 0.739 (p<0.001) in all patients. Predictive ability was similar in trauma and nontrauma patients (p=0.402) but it had a better ability to predict venous vs. arterial thromboembolism (AUROC 0.801 vs. 0.597, p<0.001). P-selectin had a similar predictive ability (AUROC 0.730; p=0.031) but no other plasma thrombotic biomarkers were significant. MA also able to discriminate between patients requiring blood product transfusion (AUROC 0.736, p<0.001) and transfusions were significantly different among patients with reduced, normal or increased MA.
Commentary: Mild coagulopathy, as seen in these patients, is common in the ICU and often leads to cessation of VTE prophylaxis given concern for bleeding. Measures of coagulation status such as platelet count, INR and PTT are relatively crude tools, and do not necessary help discriminate the true bleeding or clotting risk of an individual patient. This study of a large group of medical and surgical ICU patients with coagulopathy demonstrates a significant predictive benefit for thromboelastography in differentiating risk for thromboembolism. Utilization of TEG in addition to standard coagulation tests has the potential to help us better assess risks and benefits of coagulation therapy, often a challenging decision to make in critically ill patients. Moreover, in this study, TEG was able to accurately predict many patients with coagulopathy who would go on to require transfusions, even though the blood sampled was taken within 48 hours of admission. Thromboelastography is traditionally utilized in the OR, but this study makes a strong case for further applications in the ICU to better risk-assess our patients and ensure that we optimize care while first doing no harm.
Sara Stern-Nezer, MD, MPH
Assistant HS Professor, University of California, Irvine
Departments of Neurology & Neurosurgery